TM: I’ve been working in this industry for years and we keep trying to solve the same challenge – improving how the agents interact with customers. Is there something that all call centers share regarding this?
DA: Once you get past all the attempts to get the customer to self-serve through web self-help, on-line communities, IVRs, etc, every call center on the planet is struggling to reduce between agent variation. Some agents are empathetic, some can be downright surly. Some are very experienced; others are brand new and are not prepared. Some are hard to understand and some speak very clearly. Some are focused and listening while others are texting their friends during the call. Some are conscientious and try to do everything right and some are bored and frustrated and only do the required steps if they feel like it. Some are quick and some put you on hold numerous times. We all know this because we have all experienced it.
But these are not just anecdotal observations; there is research to back this up. A 2005 Harvard Business Review article titled "Manage Your Human Sigma" cited a study of 5,000 customer service agents, finding that the customer’s experience still depends almost entirely on the agent who takes the call. According to the study, based on post-contact interviews, the bottom ten percent of agents produced four negative encounters for every three positive ones -- that's almost a 60 percent defect rate. But perhaps even more disheartening, the top ten percent had a 14 percent defect rate. In other words, according to customers, one out of every seven encounters with the very best agents was negative.
TM: Why haven’t we made more progress on this front?
DA: The primary reason is that there is no sense of urgency about the relatively poor and always inconsistent levels of quality. CEOs don’t hold their call center leaders to the same standards they hold their manufacturing leaders to. If they did, I believe every call center leader would be summarily dismissed.
A kind of learned helplessness hangs around call centers. No one would tolerate it if the quality of their car depended on the plant it was built in. Yet, consumers, call center leaders, and every level of management tolerate the fact that the experience the customer has is a function of the agent who happens to pick up the phone. It is like seeing the long line at the post office and shrugging your shoulders; we have all become inured to poor call center service.
Related to this is what the money for capital improvement projects in call centers is really being spent on. Though no one really likes to admit it, I think we all know that the real focus in call centers is cost reduction. Some of the most popular initiatives over the last several years…off-shoring, VoIPing, online communities, agents handling multiple chats, speech-enabled IVRs…are all cost-reduction focused initiatives.
I have a friend who is a Director of Human Resources at a well known, hi-tech company. I asked him if the performance of the call centers is discussed during the business reviews. He said, “Yes, for about four minutes. The only question the SVPs ask is whether the managers will be able to keep the service costs down. They never ask about the level of CSAT or the Issue Resolution rates.”
TM: Totally agree and have seen this firsthand. Can you expand on this a little from the executive’s perspective – they get lots of reports that show improvement – what are they missing?
DA: I think the first problem is that leaders don’t know how to look at the data they are getting. There is too much use of Red-Yellow-Green in call centers today. A measure will be below standard…or Red…this month and everyone runs around wringing their hands and then it is green next month and they take their eye off it. We chase our tails every month, but no one steps back to look at the long term trends.
What leaders should be doing is taking all their output metrics…cross-sell percentage, Issue Resolution, AHT, etc…and graphing them over time on a control chart. There are only three possibilities: the measures are getting better, getting worse, or treading water.
There are few call center leaders who can put up year-over-year charts of key call center measures that show continuous improvement. That means that they are at best treading water and in many cases they are actually getting worse. If this is true then it means, de facto, that what those leaders are doing to improve their agent output measures is not working.
TM: I am surprised to hear you say that. There have been a lot of new technologies in the market to help call centers with this issue. And, we’re now sending managers to training to control the outliers. Don’t you think we have made progress?
DA: We have been trying to fix the between-agent variation problem for forty years. The industry has and continues to spend billions of dollars every year on this problem. Some try to build and maintain scripts. Some build knowledge bases and search tools for the agents to try to find answers. Some spend money on incentive plans, just trying to get agents to do what they are supposed to do. Some centers do all of these things, and of course everyone records, monitors and coaches.
We visited an offshore center the other day that tied smiley-faced balloons to the agents’ cubes, thinking that this would make them a little happier and that would make them nicer to customers. Can you imagine if they tried to improve quality at a jet engine manufacturing plant by tying yellow balloons to the tool benches? And what comes after yellow balloons…a lucky rabbit’s foot in everyone’s pocket?
One-agent-at-a-time approaches haven’t worked for forty years. If they did, we wouldn’t have the quality and between-agent variation problems we have today. Can call center leaders really look their CEOs, CFOs and CapEx committees in the eyes and with a straight face tell them that if they just spend a little more money on new recording equipment, or monitoring bells and whistles, or between call in-seat training, or improved knowledge-base search tools, or a Web-based incentive plan or some other fix-each-agent-one-at-a-time approach, then they'll see the promised improvements?
It would be the definition of insanity to think these methods are going to fix call center quality problems any time soon.
TM: But why? Why are the call center improvement approaches to reduce agent variation not working?
I think the problem here is that a flawed mental model for how to increase service quality pervades the industry. Because the industry’s improvement model is wrong, the quality never systematically improves.
The mental model goes like this: The performance of my center on any output metric like Customer Satisfaction or Issue Resolution or Compliance is the sum of the weighted averages of each agent. So, if I can improve the performance of each agent, my center-wide output measures will improve. Given this belief, most centers rely, some almost exclusively, on occasional monitoring and occasional coaching to try to improve the performance of each agent.
This dominant framework is true in theory, but not in practice. Coaching could help an individual agent improve. But given the high turnover that all centers face, center-wide metrics never improve and between-agent variation remains high because every day part of your process improvement investment is quitting and walking out the door.
TM: I read that you said the industry is "sitting on a huge mass customization opportunity." As a 20-year student of call centers, this is news to me. What are we missing?
DA: The benefits of mass customization are obvious: you give your customers more of what they value so they keep coming back to you -- that's the customization -- while raising the quality and lowering the costs of fulfillment by finding things you can standardize -- that's the mass.
Apply this thinking to the call center industry. Conventional thinking says "Oh, every incoming call is unique, there's no opportunity for standardization." Okay, but for a given type of call, large swathes of each of those calls are exactly the same. Can’t we standardize those parts and just vary what the customer wants or needs us to vary? I mean, car buyers care about color, not about bolts, so car manufacturers use standardized bolts and other sub-assemblies across models.
Look, in call centers, customers want to talk to a live agent to get their unique situation addressed, to get empathy or a solution that eases their suffering. We should absolutely deliver that. That is what they care about and want. But do they care how a required disclosure is provided to them, or how the diagnostic questions to troubleshoot their problem get asked? Of course not.
The reason it is so essential to start seeing the opportunities for mass customization is because to not see them, leaves us where we are today: each incoming call unique as a snowflake, no process defined for the agents to execute, and the call replete with sub-optimizations:
Don’t automate the greeting and customer identification steps to ensure it is branded and done correctly every time. Let’s monitor and coach them and then hope, that’s right, hope the agents remember what we told them to do and hope they decide to do it, and hope they do it right.
Don’t prerecord the disclosures; hope the agents read it word-for-word without accent issues interfering with your customers’ understanding.
Don’t use technology to ensure the right cross-sell offer is made at the right time, every time; hope your fancy variable comp plan counteracts the unyielding pressure we put on the agents to reduce their talk time.
Oh and certainly don’t error-proof the step reminding the customer to “remove any software before returning the unit” so that it can’t be skipped; we don’t need to care about that since the angry letters to the CEO get handled by another department.
TM: I’ve always been a big fan of coaching – are you saying that we’re wasting our time?
DA: Let me be clear. It is not that coaching can’t help an individual agent improve. If a particular agent wants to do a good job and you tell them how, there is a good chance they are going to try to use that information to perform better. My point is that coaching alone is not going to get us to a high level of quality. First, we have been doing it for forty years and we still have incredibly inconsistent quality. As I said, this is in part due to the typically high turnover that plagues the industry. But it is also because coaching is a weak method of process improvement because it is based on hope. You hope the agents remember what to do and do it.
TM: Okay, you’ve piqued my interest. Can you give me an example of what else we could do?
DA: Sure, if you look outside the call center industry you will find lots of them. Toyota is the absolute paragon of continuously improving quality. If you went into one of their plants and ask them how they achieve their stunning levels of quality, do you think they would tell you that the secret is that we videotape workers and coach them one-at-a-time a couple times a month? The very thought of this is laughable. If the masters of year-over-year quality improvement and cost reduction don’t rely on one-worker-at-a-time approaches to quality, you have to wonder why call center leaders have it as their go-to strategy.
What do they do instead? Their focus is, first and foremost, on continuously improving the process and then training the workers on how to execute that process.
TM: What is it about the Japanese approach, in particular Toyota, as you say, that impresses you for call centers?
DA: An easier question to answer would be, “what doesn’t impress you about Toyota’s approach?” From one small California dealership in the late 1950s, they are now the number one car manufacturer globally. Not only is this a stunning increase in market share, but also Toyota's operating margins are three times the industry average. Until the recent market carnage, Toyota's stock was up over 200 percent since the early 1990s. GM is now bankrupt.
Toyota builds great cars. How do they build great cars? They have a systematic approach for building high quality cars. They codified it in something known as the Toyota Production System. And although the call center industry has answered untold trillions of phone calls in the last 40 years, there is no documented approach for how to systematically improve call quality.
This is what the call center industry needs to take away from what Toyota has done. We have to develop a repeatable, systematic approach for improving call handling and we have to teach it to everyone in our centers.
TM: I’ve worked with a lot of companies that struggle with getting agents to read things like disclaimers and policies consistently. I can see how prerecording can eliminate some of that tug of war between agents and leaders. Tell me more about your product and how it uses the recordings to drive the process?
DA: Our agent-assisted voice product, ProtoCall, uses prerecorded audio and we have used it on millions of phone calls in domestic, near-shore and far-shore centers. It is the foundational technology that allows centers to move from an agent-centric to a process-centric approach.
We start by breaking down incoming call volume into call types. Then we assume the call is going to go in a straight line and we exactly define, just as they do in the Toyota Production System, what we want the agents to do in their systems and say to customers. We then make it easy for the agent to execute the defined process by pre-programming system actions and pre-recording audio. The result is an agent live on the call listening to the customer and navigating the call using the paths we built. If the call goes in a straight line, the agents may never have to break-in with their live voices. But if the customer asks a question that is not part of the call flow or if the customer wants to make small talk or needs a little extra hand-holding, the agent can take over the call.
TM: How does this support improvement and the reduction of between-agent variation?
DA: It helps in so many synergistic ways you can’t believe it. All the agents are following a single process. They vary it only as they need to as opposed to varying everything every time. Resources are focused on studying and improving the process first and the agents second. When a single improvement is made to the process, all the agents are instantly performing better. We can error-proof steps so that it is almost impossible for the agent to skip them. If the agents quit the performance doesn’t suffer because the expertise is baked into the pre-built process. And all this is accomplished without team meetings, laminated cards, monitoring sessions and off-phone coaching.
TM: What about the customer hearing multiple voices during the call?
DA: I understand this question, but it is also odd to me. It’s odd because the customer hears multiple voices now…there is an IVR voice and the agent’s voice and sometimes even a text-to-speech voice. Also, in exchange for all the benefits I just mentioned, wouldn’t your customers hearing one more voice be a very small price to pay?
But that aside, an approach like ours is brand new and I understand the concern. We have handled millions of phone calls and the overwhelming majority of customers don’t notice or at least never comment about hearing multiple voices. If they do notice or ask, the agent just says “I am using software to make sure the information I communicate is 100 percent accurate and easy to understand. Is that OK?” Not only do the customers not complain about this, they think it is kind of cool.
TM: Can you talk about some of the benefits you have observed using this approach?
I mentioned that customers don’t complain about hearing multiple voices. But we are not solving for customers “not complaining.” We're about making the call wildly better for all the stakeholders, and we think we have been able to do just that.
Agents love it. Why? Because we have made their jobs less stressful. They don’t have to talk as much and they know they are getting the process right. They are more relaxed and less fatigued at the end of the day. We have not seen a turnover effect yet, because we cannot get our clients to do the study, but we are sure it is there.
Customer Satisfaction has been increased because accents no longer ruin the interaction. Process adherence and compliance have both been increased to the 99.999% level because, just like they do in manufacturing, we are building the correct process and making it easy for the agent to execute it. So customers are getting the right process in a voice that is easy to understand.
TM: I’ve seen the product in action and now have a better understanding of “why” it’s going to change the way we look at call center process improvement. But you mentioned that your approach is new. When do you think the rest of the world is going to catch on?
DA: Our ProtoCall product is new -- no one is doing what we are doing and many don’t even think what we are doing is possible. It will take time. Think about ATMs, self-service kiosks at airports, and IVRs. These were also looked at warily at first but are now quotidian.
We may be starry-eyed entrepreneurs, but we feel the widespread adoption of our approach is all but assured. First, our proof-points are mounting. We already have a Telco proof-point and two financial services proof-points. Agents using our solution were dramatically superior to agents who handled the calls without our software. In addition, one of the largest technology companies in the world is piloting ProtoCall. Though we have just begun, their management is stunned at how much more effective our approach is on tough tech support phone calls.
But we were confident before these recent proof points came in. When your product is cutting the Gordian Knot on a forty-year problem, when it delights agents and results in less frustrated and more satisfied customers, when it drives dramatically better financial results while not limiting the flexibility of nuanced responses to customers, you know down to the bottom of your socks that at some point even the most risk averse, change-resistant leaders are going to have a hard time sticking with their we’ve-always-done-it-this-way approach.